GIGABYTE 9xx AM3+ thread

[rintamarotta]

I love the picture

[Lightman]

Well, I do

But I have electronics background from school. Funny description always helps understanding things.

[m411b]

200amps seems kinda high to me, as well as 100amps. I dont know of any PS that can put out that much power.

Or am I missing something?

[sin0822]

I love the picture

haha thanks

Well, I do

But I have electronics background from school. Funny description always helps understanding things.

yea i did my best with paint, and i have an EE background from the same place different location lol. Yea i try my best to help ppl understand, but its always hard to explain SMPS DC/DC to the general public, which is basically who i write too lol. So i try things like this, or make pictures with ICs in them and try to explain underneath. I all of a sudden thought hmm can i make it like a comic book? i honestly have never read a comic book, so i am not too sure if its any good. Of course we never really covered SMPS DC/DC as much as I have had to learn on my own, but its been very interesting.

200amps seems kinda high to me, as well as 100amps. I dont know of any PS that can put out that much power.

Or am I missing something?

yea hahahah 200 amps is pretty high, i mean like 200 watts, more like 250. I made that in like 10mins, b/c i am editing pics for my review. Of course this VRM can put out ca bit more than that 35A X 8 for the CPu core Imax=280A(CPU-NB is separate), but the DrMOS have to be at 300khz(switching frequency), i measured a 280khz so we are good(maybe even more current output is possible), as when SF increases current output decreases as well, that's according to the datasheet. Anyways during my testing i found that the CPu can have the VRm pull upto 260watts at 4.2ghz when i set 1.63v. I did some pretty harsh testing, i am surprised it lived. I tested out things such as claims that go the DrMOS are to blame for the droop, which is incorrect, b/c i did the mod, and the voltage increases under load, just like it does for the CPU-NB voltage stock. Then ppl said the heatsink gets hot, and i concur it gets very heated, i cannot touch it when doing extreme testing, but the DrMOS lived with hours and hours of extreme heat and no good airflow, i used the stock cooler ......

I appreciate your feedback guys, more descriptive picture it is then!

[m411b]

yea hahahah 200 amps is pretty high, i mean like 200 watts, more like 250. I made that in like 10mins, b/c i am editing pics for my review. Of course this VRM can put out ca bit more than that 35A X 8 for the CPu core Imax=280A(CPU-NB is separate), but the DrMOS have to be at 300khz(switching frequency), i measured a 280khz so we are good(maybe even more current output is possible), as when SF increases current output decreases as well, that's according to the datasheet. Anyways during my testing i found that the CPu can have the VRm pull upto 260watts at 4.2ghz when i set 1.63v. I did some pretty harsh testing, i am surprised it lived. I tested out things such as claims that go the DrMOS are to blame for the droop, which is incorrect, b/c i did the mod, and the voltage increases under load, just like it does for the CPU-NB voltage stock. Then ppl said the heatsink gets hot, and i concur it gets very heated, i cannot touch it when doing extreme testing, but the DrMOS lived with hours and hours of extreme heat and no good airflow, i used the stock cooler ......

I appreciate your feedback guys, more descriptive picture it is then!

I figured it was a typo, but wasn't quite sure. Either way thanx for the info!

[eXa]

200w at 1.4v is almost 143amps...

[Toysoldier]

Apparently you have to update the backup BIOS manually to make sure it's the same revision as the main BIOS.


To get the system Information window press F9 in the main BIOS view

To update the backup BIOS press alt+F12 when booting to get into the backup screen
Before updating my backup BIOS was at version F3.


I didn't know this and assumed that the backup BIOS was automatically updated when flashing a new BIOS
using QFlash, which is the official recommended way of flashing.
Sometimes it pays to read official guides on how to do things

[TiN_]

OMFG. Better not post that usbscope jokes in reviews. Seriosly, PLEASE.

I know, you will post anyway, but then at least set proper probe mode

[El Gappo]

[VIP3R]

at stock the BIOS is set to use the turbo function and change vcore from 1.375 to 1.475 at stock, maybe that is what you are seeing?

I have turbo turned off.

I will do some testing on stock settings and see what im getting. Probably get my DMM onto it before stressing out.

[sin0822]

OMFG. Better not post that usbscope jokes in reviews. Seriosly, PLEASE.

I know, you will post anyway, but then at least set proper probe mode

don't insult my scope, at least its not a stingray , it actually has very good specs, better than one used on sites for PSU testing, and what I am measuring at isn't even at its finest resolutions, 20mv/1ms is nothing for this scope. If you check USB scopes are almost everywhere now, they even use them in universities. BTW if you tell me not to use a USB scope in my review, you should tell every review site that uses them, and I don't know one that doesn't. Times are changing, there is no need for my to haul a rusty used POS to measure ripple.

Anyways its not something I would use for SMPS development, and it has very limited functions, no generators or anything. it does a few things and does them actually well. It can measure ripple excellent.
I like the pass fail test, i run it and it tells me how often it passes the threshold i set, it is very interesting to see max ripple spike, and then see how many times that spike occurs, the results are surprising. I can capture clear images, record videos, capture 10,000 data points whenever i feel like it. Apart from that I can run FFT tests and so on. From my previous use of desktop scope at school, this scope is just as good, if not easier to use once you read the software manual. I also don't have to worry about isolating the ground. It might not be a $1 tektronics unit, but its good enough. But wait how about I use the same timescale for all my testing......... i looked at your review a bit closer this time Tin...

its in AC coupling, what are you talking about? you want me to measure ripple in DC mode? LOL





I am using the same timescales you used, and ill test the UD7, and if my results are like yours then nothing is wrong, but I am not going to go measure most at 1ms and then switch to 40ns for one.

[TiN_]

No comments.

If you so sure about very good specs of USB scope (there are few of them which are good, but they are priced much more than few hundreds of bucs, and still limited compared to normal desktop scope), why you even bother me with PM asking what is wrong, and how to make it right?

I don't owe you anything, nor other reviewers sites, they can use any devices they want, even ancient plotters if they want to. Somehow I wanted to see your reviews better, that's my fault

But wait how about I use the same timescale for all my testing......... i looked at your review a bit closer this time Tin...

I will not help you this time, take care of yourself, you have EE background in university, right? It's very basics.
Now you see numbers and nice graphics, and do wierd buble-speach images, but you still unable to see what that numbers mean.
Sorry for being wrong, scope is not a problem in case of wrong meas setup, no matter if it's done by USB or 50k$ LeCroy, it will be same wrong. Continue please, let other EE have fun

[sin0822]

i am sorry? what you want me to do it at 10x instead of 1x so the load on the PSU wont be as much, don't worry Tin. I know there is a lot of noise on those results, besides under transients during wPrime i did end up with about 110mv Vpp, i am not fooling anyone. And FYi i know how to use a scope very well, please don't insult, those were not my final images. You enjoy trying to discredit me, don't you? i can add smile faces as well when I feel like being .

I understand that 10x has great impedance on most probes than 1x. But I have a problem with using 10x, and that problem is the signal I am trying to measure is super small, and 10x decreases my accuracy 10x. So its give and take, i am working on trying to find a better probe(one at 1x and with very high impedance), but for the time being, I will show 10x compensated and 1x compensated, and i know how to compensate a BNC, and yea my USB "junk" as you call it has a 1khz compensation signal

Anyways Tin, at 1x i could be loading the output i am measuring, as 10x is much moer typcial and loads my circuit less. But i am loading 100-200watts, i doubt it'd make much difference, so Ill do a little test.

[sin0822]

over here some guy posted the BIOS F4C, said it improved his vdroop a bit.
http://www.overclock.net/14310568-post604.html

[Fuse]

Hey there, can someone send me this beta BIOS F3b for GA-990FXA-UD3 please?
Thanks.

[m411b]

over here some guy posted the BIOS F4C, said it improved his vdroop a bit.
http://www.overclock.net/14310568-post604.html

This BIOS didn't change a thing for me! Not even worth the time to flash it!!

[sin0822]

yea I haven't had time to try it out. Hopefully we will have a F5 beta soon as well as the final F4 BIOS.

[Toysoldier]

New BIOS for the UD5 ... : http://forum.giga-byte.co.uk/upload/...XA-UD5_F5c.zip

I have no idea what's new or changed. Running with it now. No change in vdroop

Will look for new options in the BIOS.

[Augustus]

Hey there, can someone send me this beta BIOS F3b for GA-990FXA-UD3 please?
Thanks.

Here you go mate : http://www.overclock.net/amd-motherb...l#post14400432

[Chainlink]

Hi all. Long time reader / first post. Just upgraded to a GB 990FX UD5 (F5C bios) from a MSI 790FX GD70 in anticipation of bullzoder. Running a 1055t (takes 1.5v+ to hit 4ghz, not a good chip).

I heard of vdroop issues with the new gb 990fx boards but I didn't think it would be anything like what I'm experiencing. Bios setting is +.2v (stock is 1.4) so effective 1.6v. Initial idle in windows is 1.648v, load in p95 is 1.552-1.564, during 3d its around 1.62, then later idle is 1.472-1.488. I really don't understand WTF is going on with this board. I'm not super concerned with the voltage with this chip, max temp on p95 is 44c, but I am concerned about how this could affect BD.

System is stable but these voltages are crazy. Also, first boot bios sets stock voltage to 1.475 (should be 1.4). After entering windows, restarting, entering bios, default voltage drops to the correct 1.4. WTF is that? Comments/concerns/ideas?

[PatRaceTin]

my friend also mess up with those droop on UD7

[eXa]

Just built a new computer for my GF with the 990FX-UD3, now i need the new bios since F2 bugs out on boot if i try to enter bios.
Already installed windows with ACHI on a crucial M4 128GB so i dont want to disable achi either.

[sin0822]

I believe you have turbo core on or some C1E or some type of power saving for idle to drop to 1.47.

Hi all. Long time reader / first post. Just upgraded to a GB 990FX UD5 (F5C bios) from a MSI 790FX GD70 in anticipation of bullzoder. Running a 1055t (takes 1.5v+ to hit 4ghz, not a good chip).

I heard of vdroop issues with the new gb 990fx boards but I didn't think it would be anything like what I'm experiencing. Bios setting is +.2v (stock is 1.4) so effective 1.6v. Initial idle in windows is 1.648v, load in p95 is 1.552-1.564, during 3d its around 1.62, then later idle is 1.472-1.488. I really don't understand WTF is going on with this board. I'm not super concerned with the voltage with this chip, max temp on p95 is 44c, but I am concerned about how this could affect BD.

System is stable but these voltages are crazy. Also, first boot bios sets stock voltage to 1.475 (should be 1.4). After entering windows, restarting, entering bios, default voltage drops to the correct 1.4. WTF is that? Comments/concerns/ideas?

[Chainlink]

I believe you have turbo core on or some C1E or some type of power saving for idle to drop to 1.47.

Indeed it was actually cool n quiet causing the drop at idle 1.47. However, it seems to keep the voltage reasonable at idle. Only thing is I have to hit it with a load before it takes effect. I think I'll mess around and find out where I can set voltage so 3d is stable, and keep cnq enabled.

[eXa]

Ah, finally found the F3B bios, will update to it in a couple of hours.

Edit: Installed, now it boots and enters bios without problems